Objectives. Animal assisted therapy (AAT), also termed Pet-Therapy, may be beneficial particularly to older people with cognitive disturbances. Efficacy evaluation of AAT in cognitive impairment is frequently biased by poor methodological design, and conclusive evidence is still lacking. Aim of this study was to verify the clinical effect of a medium term dog-AAT in non-hospitalised patients affected by mild cognitive impairment due to Alzheimer's disease (AD) versus a control group. Methods. The controlled study included out-patients diagnosed with early stage or mild AD and centrally randomised into two groups: AAT treatment with dogs and Controls (C) both followed for eight months. At baseline and after 8 months, Barthel index (functional status), ADAS test (AD Assessment Scale for cognitive status) and Cornell Scale (depression condition) were assessed. The statistical analysis was performed by the Mann-Whitney U-test for non-parametric independent data. Results. Twelve patients for each group (aged 82.1 and 78.3 years in AAT and Controls) were evaluable at 8 months. In AAT group, Barthel index showed less deterioration in daily life activities, ADAS test indicated a slowing of the progression of AD, and Cornell scale showed the absence of a depressive syndrome. Discussion. This study confirms the feasibility of AAT with dogs in low-mild AD. AAT lead to a lower impairment of daily life activities, and cognitive deterioration. An extensive multi-centre controlled study is required to reach a higher statistical power, even if objective difficulties of keeping homogeneity (dog selection, characteristics of activities, etc.) make such a project unfeasible.